Modeling 2,4-D Transport in Turfgrass, Thatch and Soil - American

Chapter 11

Downloaded by DUKE UNIV on June 26, 2012 | http://pubs.acs.org Publication Date (Web): December 20, 2009 | doi: 10.1021/bk-2009-1028.ch011

Modeling 2,4-D Transport in Turfgrass, Thatch and Soil S. Raturi, R.L. Hill* and M.J. Carroll Department of Natural ResourceSciences and Landscape Architecture University of Maryland, College Park, MD 20742

The transport of 2,4-D [(2,4-dichlorophenoxy) acetic acid] was measured for replicated soil columns containing a surface layer of turfgrass thatch and for soil columns devoid of thatch. Following the application of bromide to determine transport parameters, 2,4-D was surface-applied to undisturbed columns under steady state unsaturated conditions. Linear equilibrium (LEM), two-site non-equilibrium (2SNE) and one-site kinetic non-equilibrium (1SNE) models were curve-fitted to experimentally determined breakthrough curves. Modeling of bromide transport did not present strong evidence of significant two domain flow. All models provided reasonable estimates of 2,4-D transport, with slightly improved fits from the 2SNE model when the retardation factor was a fitting parameter. When retardation factors based on laboratorymeasured adsorption coefficients were used, significantly improved fits from the 2SNE model were obtained in comparison to the LEM and 1SNE models, suggesting the occurrence of both instantaneous and kinetically driven adsorption. Parameter estimations of 2,4-D retardation factors based solely on curve fitting techniques may result in inappropriate model selection, although excellent curve fit solutions during model calibration may have been previously obtained.

© 2009 American Chemical Society

157

In Turf Grass: Pesticide Exposure Assessment and Predictive Modeling Tools; Nett, M., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2010.

158


Introduction Evaluation of the potential for pesticides to contaminate groundwater requires an understanding of the transport mechanisms that occur in the field to accurately represent these mechanisms in simulation models. The presence of thatch complicates prediction of pesticide transport in turf systems, since surface-applied pesticides must pass through an organically rich thatch layer prior to entering the soil. The linear equilibrium (LEM), the two-site kinetic non-equilibrium (2SNE), and the one-site kinetic non-equilibrium (1SNE) models are process based forms of the convective dispersive equation (CDE) used to describe pesticide transport within soils. Alternate forms of this equation may be used to describe single or two domain physical flow phenomena or to describe the contribution of instantaneous or kinetically driven adsorption during the transport process. Thatch has a pore space arrangement similar to that of a course sand and a chemical composition resembling a young organic soil (1). The high organic matter content of thatch allows this medium to readily sorb non-polar compounds (2, 3, 4). The rapid drainage properties of this medium result in short solution residence times, which can minimize the sorption of ionic compounds to thatch (5, 6). This suggests that solute transport models which use non-equilibrium or two-site sorption may be better able to predict pesticide transport in thatched turf than an LEM. Convective dispersive equation based models use a retardation factor (R) to account for pesticide sorption. Several approaches have been used to obtain R for model simulations (7, 8, 9); most, however, involve some form of curve fitting to the data being modeled. Obtaining R in such a manner invariably improves model performance, but leaves open to interpretation the true nature of pesticide transport sorption dynamics. Retardation factors based on independently measured sorption coefficients using modified batch flow techniques should be more appropriate than using retardation factors derived as simple optimized fitting parameters, since there is a physical basis for the retardation factor which is based on the experimental conditions. The objectives of this study were (1) to evaluate the effects of thatch on 2,4D transport through undisturbed soil columns, (2) to compare the use of LEM, 2SNE and 1SNE models to predict 2,4-D transport through the soil columns containing a surface layer of thatch and columns devoid of thatch and (3) to evaluate the effectiveness of retardation factors based on laboratory measured adsorption coefficients and model-fitted retardation factors to simulate 2,4-D transport.

Theoretical Background The one dimensional convective dispersive equation (CDE) for steady-state transport of a solute through homogeneous soil is (10): R*δC/δt = D*(δ2C/δx2) - v* (δC/δx)


159 -3

Where C is solution phase solute concentration (μg cm ), t is time (h), D is the hydrodynamic dispersion coefficient (cm2 h-1), R is the retardation factor (dimensionless), x is distance from solute origin (cm) and v is the average pore water velocity (cm h-1). The R term reduces to one for non-reactive solutes and is greater than 1 when solute retention occurs. The retardation factor is defined as (11):


R = 1+[ρKf (1/n)C (1/n-1) /θ where ρ is the soil bulk density (g cm-3), θ is the volumetric water content (cm3 cm-3) and Kf and 1/n are Freundlich empirical distribution coefficient constants that characterize sorption. The simplest approach is to assume that all pesticide sorption sites are identical, and that equilibrium occurs instantaneously between the pesticide in the bulk soil solution and the pesticide adsorbed. This mathematical approach is called linear equilibrium sorption. Where bimodal porosity leads to two-region flow, or situations where the sorption process is controlled by two-site kinetic non-equilibrium sorption processes, non-equilibrium models may more accurately describe the transport of pesticides through soil. Chemical nonequilibrium models consider adsorption on some of the sorption sites to be instantaneous, while sorption on the remaining sites is governed by first order kinetics (12). The two-site chemical non-equilibrium model (2SNE) conceptually divides the porous medium into two sorption sites: type-1 sites assume equilibrium sorption and type-2 sites assume sorption processes as a first-order kinetic reaction (13). In contrast, physical non-equilibrium is often modeled by using a two-region dual porosity type formulation. The two-region transport model assumes the liquid phase can be partitioned into mobile (flowing) and immobile (stagnant) regions. Solute exchange between the two liquid regions is modeled as a first-order process. The concepts are different for both chemical and physical non-equilibrium CDE, however, they can be put into the same dimensionless form (2SNE) for conditions of linear adsorption and steady-state water flow (14): β*R*δC1/δt = 1/P*(δ2C1/δx2) - δC1/δx - ω(C1-C2) – μ1C1 + γ1(x) (1-β)*R*δC2/δt = ω(C1 - C2) - μ2C2 + γ2(x) Where the subscripts 1 and 2 refer to equilibrium and non-equilibrium sites, respectively, β is a partitioning coefficient, ω is a dimensionless mass transfer coefficient, P is the Peclet number and μ (h-1) and γ (ug h -1) define first-order decay and zero-order production terms, respectively, each represented in component contributions of both the liquid and solid phases. Customarily, β and ω are obtained by fitting solute BTC’s to the nonequilibrium model using a non-linear least squares minimization technique (15). The values of β and ω obtained from the BTC’s of non-interacting solutes can be used to evaluate the potential contributions from two-region flow. In the absence of two-region flow, β and ω may be used to evaluate the contributions from two-site kinetic non-equilibrium sorption (16). For interacting solutes, β


160


represents the fraction of instantaneous solute retardation in the two-site nonequilibrium model, and ω the ratio of hydrodynamic residence time to characteristic time for sorption. They are equivalent to β = (θ +f ρK)/(θ + ρK) and ω = k2 (1- β)RL/v, where f is the fraction of equilibrium-type sorption sites, K is Kf when 1/n is unity, L is the length of transport (cm) and k2 is the desorption rate constant (h-1). The one-site non-equilibrium model is a special case of the two-site nonequilibrium model. A one-site model assumes that sorption of the pesticide is kinetically driven (type-2 sites), thus the fraction of type-1 sites (f) is reduced to zero (17). The dimensionless coefficients used for the one-site model are the same as those used for the two-site model except that β is defined as β = 1/R, and ω as ω = k2 (R- 1)L/v (17).

Materials and Methods Sample Collection Soil and turfgrass thatch were collected from two sites at the University of Maryland Turfgrass Research and Education Facility in Silver Spring, Maryland. One site was a three and half year old stand of Southshore creeping bentgrass (Agrostis stolonifera), and the other a six year old stand of Meyer Zoysiagrass (Zoysia japonica Steud.). Visual inspection of the bentgrass site revealed the presence of a finely granulated 1.5 to 2.0 cm thick thatch layer. The zoysiagrass site contained a 3.0 to 3.5 cm thick thatch layer that consisted primarily of non-decomposed and partially decomposed rhizomes, stolons and tillers. The soil at the zoysiagrass site was classified as a Sassafras loamy sand (fine loamy, mixed, mesic, Typic Hapludult; 81.2% sand, 10.2% silt, and 8.7% clay) whereas the soil at the bentgrass site was classified as a Sassafras sandy loam (fine sandy, mixed, mesic, Typic Hapludult; 71.2% sand, 15.8% silt, and 12.8% clay). The saturated soil hydraulic conductivity was 24.4 cm h-1 at the zoysiagrass site and 18.2 cm h-1 at the bentgrass site. The thatch and soil used to determine 2,4-D sorption isotherms were collected by removing the thatch using a sod cutter. Prior to using the sod cutter, all verdure was removed by scalping the turf with a walk-behind greens mower. The intact rolls of the turfgrass thatch were shredded using a modified wood chipper and the shredded field moist material passed through a 4 mm screen. The soil directly beneath the thatch (2 cm depth) was collected using a shovel, and the field moist soil passed through a 4 mm screen. The soil columns used in the leaching study were extracted from the surface of each site using a specially designed drop hammer-sleeve assembly. Four soil and four soil plus thatch columns, 12 cm length by 10 cm diameter, were collected from each site. The columns containing soil only were obtained after using a shovel to remove all above ground thatch and foliage. The columns were brought to the laboratory and saturated from the bottom immediately after collection.


161


Sorption Isotherms A modified batch/flow technique was used to measure pesticide sorption (5). The technique involved the use of a mechanical vacuum extractor. This device controls the rate at which a solution moves through a column of thatch or soil. The columns were created by packing known amounts of media into syringe tube barrels after placing a sheet of glass fiber filter paper (Fisher Scientific, Pittsburgh, PA, Cat. No: 09-804-70C) into the bottom of each barrel. Since the sample was not shaken during the procedure, little disruption of the medium aggregates and organic matter occurred. Moreover, the flowing conditions used in this modified batch/flow technique better represent the physiochemical interactions that occur in the field. A combination of technical grade 2,4-D and ring-labeled 14C 2,4-D were used to determine the sorption of 2,4-D to thatch and soil. Sorption isotherms were determined by leaching 30 mL of 1, 10, 30, or 100 mg 2,4-D L-1 through samples of thatch and soil for 24 hours. All four solutions contained 2.31 x 105 Bq L-1, of 14C 2,4-D. The radioactivity of 1 mL of leachate plus 5 mL of scintillation cocktail was determined by liquid scintillation counting (LSC). Sorption of 2,4-D to any material other than thatch or soil was accounted for by including syringe tube blanks. The blanks were identical to the syringe tubes containing thatch or soil, except they contained no thatch or soil. Sorption of 2,4-D at 24 h was fitted to the linear form of the Freundlich equation. Regression analyses were used to calculate the capacity (Kf) and intensity (1/n) of sorption to each medium. Student's t-tests were used to test for homogeneity of slopes and to compare equation intercepts. Leaching Experiment The bottom end of the columns obtained from the field were placed into separate funnels fitted with a rubber o-ring and a 12-µm pore diameter saturated, porous, stainless steel plate. The columns were made vacuum tight to each funnel using adhesive acrylic caulking and the funnel inserted into one port of a multi-port vacuum chamber. A null balance vacuum regulator was used to maintain a constant pressure of -10 kPa within each vacuum chamber. A 0.001 M CaCl2 solution was continuously applied to each column using a specially designed drop emitter that uniformly distributed the solution to the surface of each column (modified design of 18). Leachate was collected in sterile plastic cups beneath the funnel of each column within the vacuum chamber. Once steady-state flow (0.85 cm h-1) was achieved in all columns, 10 mL of 300 mg bromide L-1 (KBr salt) was surface-applied uniformly to each column. Leachate was then collected every 30 minutes for the next 12 hours. The concentration of Br- in the leachate was determined using standard ion chromatography techniques outlined in the Dionex Users Guide. After the initial leaching period, 10 mL of 88 mg 2,4-D L-1 was uniformly surface-applied to each column. The 2,4-D solution contained 2.31 x 105 Bq L-1 of ring-labeled 14C 2,4-D. The addition to each column was equivalent to a field rate of 1.12 kg 2,4-D ha-1. After adding 2,4-D, the leaching solution inputs and



162 vacuum applied to the base of each column were discontinued for 24 hours to permit sorption of 2,4-D to the thatch and soil. During this time, all columns were covered with plastic wrap to prevent volatile losses of 2,4-D. After the 24 h adsorption period, the plastic wrap was removed, the emitters placed back atop each column, and the vacuum engaged. Leachate was then collected every half an hour for the next 18 h with 2,4-D in the leachate being determined by LSC as previously described. To verify that the radioactivity measured by LSC was 14C 2,4-D and not one of its primary metabolites, every 4 h during the leaching event 1mL subsamples of leachate were collected from a single column for each of the four column treatments. A 25 cm x 4.6 mm ID 5µm Supelcosil LC-18 (Supelco p/n 5-8298, Bellefonte, PA) column installed into an Hewlett Packard (HP) model 1050 liquid chromatographic system equipped with a quartenary pumping module, automatic liquid sampler and HP model 79853A variable wavelength UV detector was used to determine the concentration of 2,4-D in these samples. Analytical standards of 2,4-D were analyzed concurrently with the leachate samples to confirm the accuracy of the high performance liquid chromatography (HPLC) analysis. The limit of quantification was 0.02 µg 2,4-D mL-1. After collecting the last leachate sample, the columns were removed from the vacuum chamber and sectioned into halves. In the columns containing thatch, the thickness of the thatch layer was measured before separating the thatch and soil. One half of each core section was used to determine the water content in the section. The other half of the section was immediately frozen for later determination of the amount of 2,4-D present in the section. At a later date, the frozen section was thawed and shaken for 2 hours in a 50:50 water and methanol solution. The resulting slurry was then subjected to vacuum filtration and the filtrate analyzed for 14C. The amount of 14C remaining in the sample was determined through combustion using a biological material oxidizer with the amount of 14C evolved measured by LSC. Estimating Transport Parameters From Breakthrough Curves (BTC’s) Convective transport parameters were estimated by a least squares minimization procedure (CXTFIT, 19) using the bromide breakthrough data. All CXTFIT calculations were performed under flux type boundary conditions. Actual mean pore water velocities were used and the retardation factor, R, was assumed to be equal to 1. One and two domain flow forms of the convective dispersive equation were curve-fit to the bromide leachate data. Values of the dispersion coefficient were used in subsequent 2,4-D simulations. The LEM model was fitted to the 2,4-D transport data using CXTFIT. The 2SNE and 1SNE models were fitted using CXT4 (13). All models used calculated mean pore water velocities and the bromide-fitted dispersion coefficients. Degradation coefficients (µ) for the LEM model were estimated by applying an exponential decay function to mass balance quantities. The dimensional degradation coefficients (Ψ = µL/v; 20) in the 2SNE and 1SNE models were calculated from the appropriate µ values. The retardation factors were calculated based on the column measured values of θ, ρ, maximum pesticide breakthrough concentrations and 2,4-D adsorption coefficients (Kf and



163 1/n). Pesticide retardation factors for individual columns were calculated using thatch and soil Kfs in a volume-averaged approach where the relative length of the thatch and soil layers were used as weighing factors in calculating a mean retardation factor for each column. Pulse is the duration of solute addition and was a fitting parameter during all model simulations. The dimensionless partitioning coefficient (β), and the dimensionless rate coefficient (ω) which specify the degree of either chemical or physical non-equilibrium were fitted for the 2SNE model. For the 1SNE model, the value of β was calculated as β= 1/R. The value of ω and pulse were fitted in the 1SNE model. Simulations were repeated for all columns a second time, with retardation factors being fitted so comparisons could be made of model fits using measured and fitted retardation factors. The value of R determined from the two-site model was assumed to be the same for the 1SNE model. The value of β was then calculated as β = 1/R. The value of ω and pulse were fitted in the 1SNE model simulations. Simulations were repeated for all columns a third and fourth time, and the previously stated methods were followed in each case except the degradation term was assumed to be zero.

Results and Discussion Freundlich sorption parameters for 2,4-D in thatch and soil are presented in Table I. The sorption of 2,4-D to thatch was greater than to soil. The 2,4-D sorption capacity of the two turfgrass species were similar, and are represented by a single set of Freundlich parameters. Table I. Freundlich Sorption Parameters for 2,4-D in Bentgrass and Zoysiagrass Thatch and the Soil Residing Below Each Thatch Layer Media Thatchb Bentgrass Site Soil Zoysiagrass Site Soil a

(1-1/n) 1/n

Kfa

1/n

r2

3.14xd

0.86 (± 0.01)x

0.99

-0.15 (± 0.04)

0.71y

0.86 (± 0.03)x

0.98

-0.46 (± 0.06)

0.35z

0.83 (± 0.04)x

0.96

log Kf 0.50 (± 0.02)c

-1

Kf = mg L kg There was no difference in the 2,4-D sorption capacity of the two turfgrass species thatch thus a single Freundlich isotherm was used to describe the sorption of 2,4-D to turfgrass species thatch. c Values in the parenthesis indicate standard errors of estimates. d Values followed by the same letter in a column are not significantly different (P≤ 0.05). b


164


Leaching The physical properties and transport conditions for the soil and thatch plus soil columns are summarized in Table II. The soil water contents were slightly greater for the columns containing a surface layer of thatch compared to the columns devoid of thatch. The presence of thatch also decreased the mean bulk density of the thatch plus soil columns compared to the columns devoid of thatch. The amount of 2,4-D determined by LSC analysis was plotted against the amount determined by HPLC analysis for the columns where analysis by both analytical techniques took place. There was good agreement between the two techniques (r2 >0.95) indicating that the LSC data represented the presence of 14 C 2,4-D and not one or more of 2,4-D’s metabolites. Total mass LSC recoveries of 2,4-D within the columns were 81.18 (±2.35)% for the zoysiagrass thatch+soil, 72.26 (±3.02)% for the zoysiagrass soil, 84.91 (±8.17)% for the bentgrass thatch + soil, and 90.86 (±7.03)% for the bentgrass soil. The bentgrass columns devoid of thatch had the greatest 2,4-D leaching losses (43.11±1.10%). Conversely, columns having a surface layer of bentgrass thatch had the lowest 2,4-D leaching losses for the four column types examined (17.45 ± 1.83%). Columns having a 3.5 year old, 1.7 cm surface layer of bentgrass thatch were more effective (P=0.0078) in reducing 2,4-D leaching than columns having a 6 year old, 3.2-cm surface layer of zoysiagrass thatch (29.03 ± 3.01%). There was no difference (P=0.162) in the amount of 2,4-D leached from the zoysiagrass columns devoid of thatch (34.35 ± 2.04%) than from the zoysiagrass columns containing thatch. The coarser nature of the zoysiagrass thatch may have limited the amount of 2,4-D that was initially intercepted, compared to the more tightly intertwined bentgrass thatch. The leaching results demonstrate that the turfgrass species from which a mature thatch layer originates can have a greater influence on 2,4-D attenuation than does the age or thickness of the thatch layer. Model Evaluation Both one and two domain flow models did well in estimating bromide transport with reasonable estimations of transport parameters obtained for all columns. Peak bromide concentrations occurred prior to the leaching of one pore volume in some columns which might suggest two domain flow, but both models performed well and gave close fits of the peak concentrations with r2 values of 0.98 to 0.99 obtained for most columns. Since both the one and two domain


165 Table II. Mean Physical Properties and Experimental Parameters for Columns Containing a Surface Layer of Thatch and Columns Devoid of Thatch Used in the 2,4-D Leaching Study Mean Pore Water Velocity cm h-1

Darcy Flux cm h-1

Soil Water Content cm3cm-3

Bulk Density g cm-3

Zoysiagrass Thatch+Soil

2.41

0.86

0.34

1.30

Zoysiagrass Site Soil

2.60

0.83

0.32

1.66

Bentgrass Thatch+Soil

2.77

0.87

0.31

1.24

Bentgrass Site Soil

3.43

0.88

0.25

1.54


Column ID

flow models resulted in good fits to the measured bromide leaching, there is not strong evidence that significant amounts of two domain flow was occurring. Mass balance derived 2,4-D degradation coefficients (µ) values for columns containing the zoysiagrass thatch and bentgrass thatch were 0.014 and 0.012 h-1, respectively. Mass balance derived 2,4-D µ values for zoysiagrass and bentgrass site soil columns were 0.034 and 0.007 h-1, respectively. When µ and Ψ (Ψ = µL/v) for the appropriate models were included in the model simulations, there were no perceivable improvements in the quality of the fits compared to model simulations where a degradation term was not included. The solute leaching concentrations comparing the liquid scintillation counting and HPLC methodologies also indicated that there was minimal degradation of 2,4D in the leachate. Thus, model simulation values without the use of degradation coefficients are presented and discussed (Tables III and IV). If model evaluation is based on the coefficient of determination, all three models described 2,4-D transport fairly well, with slightly improved fits resulting from the 2SNE model when R was a fitting parameter. This overall quality of fits may be partially attributable to volume averaging thatch physical properties over the column length when calculating transport parameters. The Damköhler number (ω), which is a ratio of hydraulic residence time to reaction time and, as such, characterizes the degree of non-equilibrium, is often used as a criterion for linear equilibrium model validity (21, 22). Results of the sensitivity analysis showed that values of ω >100 were generally indicative of LEM validity and lack of significant transport non-equilibrium (16, 23). The fact that ω values were greater than 100 for columns containing the surface layer of thatch suggested that the LEM model may have been appropriate for describing 2,4-D transport. However, when R was based on a laboratory measured sorption coefficient, the 2SNE model gave significantly improved fits, indicating two-site non-equilibrium adsorption may have occurred. Similar results of two-site sorption non-equilibrium exhibited by 2,4-D were also reported by Khan (24)



166 and Rao et al., (25). They attributed sorption non-equilibrium of 2,4-D to the rate-limited interaction between 2,4-D and the sorbent organic matter (24, 26). The fitted retardation coefficients were 37 to 75% lower than the measured retardation factors for the LEM, and 35 to 68% lower for the 2SNE models. Model fits which were based on measured R values also gave more realistic values of β than model fits where R was a fitting parameter (Rfit). When R was a fitting parameter with the 2SNE model, mean values of β increased for the bentgrass thatch plus soil columns from 0.41 to 0.84 (108% increase), and increased for the zoysiagrass thatch plus soil columns from 0.42 to 0.82 (96% increase). Retardation factors based on laboratory measured sorption coefficients (Rmes) should be more appropriate than using retardation factors derived as a simple optimized fitting parameter, since there is a physical basis for the R value which is based on the experimental conditions. Using estimated R (Rfit) values in prediction models may underestimate or overestimate subsequent 2,4-D transport and may not accurately represent the physical processes occurring during 2,4-D transport, since the model is optimally fitting for the parameters. Similar conclusions were also obtained by Brusseau (27), where the utility of fitting a model to measured laboratory and field experiment data to model evaluation and data analysis was examined. They reported that the misuse of calibration can lead to a mistaken belief that the model accurately represents the physical system which can result in a misinterpretation of the factors controlling solute transport. Non-equilibrium parameters (β and ω) for 2,4-D transport were optimized by fitting 2,4-D BTCs to the 2SNE model using independent estimates of R and v (Tables III and IV). Because bromide exhibited no significant two-domain flow in the columns; β and ω values obtained for the 2,4-D BTCs can be interpreted primarily as sorption related non-equilibrium parameters (16, 19). In this case, calculated values of f (fraction of sorbent for which sorption is instantaneous) and k2 (the desorption rate constant) using β and ω terms, respectively, may be interpreted as relating to sorption non-equilibrium (2SNE) without confounding effects from two-region or transport related nonequilibrium (28). Values of f were 0.19, 0.14, 0.11 and 0.30 for columns containing a surface layer of bentgrass thatch, zoysiagrass thatch, bentgrass site soil columns and zoysiagrass site soil columns, respectively. These f values indicated that a significant fraction of the sorption sites did not participate in instantaneous retardation in these columns during 2,4-D transport. Values of k2 for columns containing a surface layer of bentgrass thatch, zoysiagrass thatch,


167


Table III. Transport Parameters for 2,4-D Breakthrough Curves From the Linear Equilibrium (LEM), Two-Site Non-Equilibrium (2SNE) and OneSite Kinetic Non-Equilibrium (1SNE) Models for Zoysiagrass Thatch+Soil Columns (ZT) and Soil Columns Devoid of Thatch (ZS) Using Fitted (ZTfit, ZSfit) and Measured (ZTmes, ZSmes) Retardation Factors Column ID

Model

v

D

Rmes

Rfit

β

ω

r2

ZTmes

LEM

2.77

3.45

4.05

--

--

--

0.15

2SNE

2.77

3.45

4.05

--

0.419

0.218

0.93

1SNE

2.77

3.45

4.05

--

0.247

2.62

0.31

LEM

2.77

3.45

--

1.71

--

--

0.93

2SNE

2.77

3.45

--

1.72

0.822

1115

0.94

1SNE

2.77

3.45

--

1.72

0.582

385

0.94

LEM

3.43

2.25

3.19

--

--

--

0.05

2SNE

3.43

2.25

3.19

--

0.523

0.547

0.92

1SNE

3.43

2.25

3.19

--

0.314

2.41

0.66

LEM

3.43

2.25

--

1.74

--

--

0.93

2SNE

3.43

2.25

--

1.83

0.464

7.40

0.97

1SNE

3.43

2.25

--

1.83

0.553

5.12

0.96

ZTfit

ZSmes

ZSfit

bentgrass site soil columns and zoysiagrass site soil columns were 0.03, 0.02, 0.06 and 0.12, respectively, indicating that there were relatively large differences in 2,4-D desorption in the columns containing a surface layer of thatch and columns devoid of thatch.


168


Table IV. Transport Parameters for 2,4-D Breakthrough Curves From the Linear Equilibrium (LEM), Two-Site Non-Equilibrium (2SNE) and OneSite Kinetic Non-Equilibrium (1SNE) Models for Bentgrass Thatch+Soil Columns (BT) and Soil Columns Devoid of Thatch (BS) Using Fitted (BTfit , BSfit ) and Measured (BTmes , BSmes) Retardation Factors Column ID

Model

v

D

Rmes

Rfit

β

ω

r2

BTmes

LEM

2.41

6.87

4.77

--

--

--

0.09

2SNE

2.41

6.87

4.77

--

0.405

0.316

0.78

1SNE

2.41

6.87

4.77

--

0.212

4.62

0.27

LEM

2.41

6.87

--

1.88

--

--

0.78

2SNE

2.41

6.87

--

1.91

0.844

1810

0.81

1SNE

2.41

6.87

--

1.91

0.528

248

0.79

LEM

2.61

4.61

4.75

--

--

--

0.35

2SNE

2.61

4.61

4.75

--

0.299

0.951

0.89

1SNE

2.61

4.61

4.75

--

0.214

2.26

0.49

LEM

2.61

4.61

--

1.51

--

--

0.87

2SNE

2.61

4.61

--

1.79

0.633

4.99

0.94

1SNE

2.61

4.61

--

1.79

0.564

24.61

0.91

BTfit

BSmes

BSfit

Conclusions The presence of bentgrass thatch reduced the leaching of 2,4-D applied to turfgrass. When 2,4-D breakthrough curves were fitted to the different forms of the convective dispersive equation, all models provided reasonable estimates of 2,4-D transport when the retardation factor was a fitting parameter. When retardation factors derived from laboratory determined sorption coefficients were used, significantly improved fits from the 2SNE model were obtained in comparison to the LEM and 1SNE models, indicating the occurrence of both instantaneous and kinetically driven adsorption. Retardation factors derived from laboratory determined adsorption coefficients provided a more realistic estimation of processes involved in 2,4-D transport. Mass balance-derived degradation coefficients did not result in improved model estimations, and had limited utility because of the minimal quantities of 2,4-D degradation observed. While there were differences in the amount of 2,4-D leached from columns containing the thatch and those devoid of thatch, the presence of thatch did not


169


affect model performance. Column k2 values revealed that 2,4-D was more tightly sorbed to thatch than soil, however, the presence of thatch did not appear to alter the fraction of sorption sites associated with instantaneous sorption within the columns. The proportion of instantaneous sorption sites was relatively low in all columns, which may explain why non-equilibrium transport of 2,4-D was observed in all columns. More importantly, the results of this study showed that parameter estimations of 2,4-D retardation factors based solely on curve-fitting techniques may result in inappropriate model selection, even though excellent curve fit solutions during model calibration may have been previously obtained.

Acknowledgements The authors express their appreciation to the United States Golf Association Green Section Research,The Maryland Turfgrass Council and Maryland Agricultural Experiment Station for providing financial assistance to conduct this research. Special thanks are extended to Dr. A. Herner (now retired) and Emy Pheil at the Environmental Chemistry Laboratory, USDA, Beltsville for their assistance during this research.

References 1. Hurto, K.A., Turgeon, A.J. and Spomer, L.A. Agron. J. 1980, 72, 165-167. 2. Hurto, K.A. and Turgeon, A.J. Weed Sci. 1979, 27, 141-146. 3. Dell, C.J., Throssell, C.S., Bischoff, M. and Turco, R.F. J. Environ. Qual. 1994, 23, 92-96. 4. Lickfeldt, D.W. and Branham, B.E. J. Environ. Qual. 1995, 24, 980-985. 5. Raturi, S., Carroll, M.J., Hill, R.L., Pfeil, E. and Herner. A.E. Intern. Turf. Soc. Res. J. 1997, 8, 187-196. 6. Carroll, M.J, and R.L. Hill United State Golf Association Turfgrass and Environmental Research Onlin 2003, 2:(5) 1-9. (ISSN 1541-0277, http//www.usgatero.msu.edu. TGIF Record Number 85740). 7. Roberts, P.V., Goltz, M.N. and Mackay, D.M. Water Resour. Res. 1986, 22, 2047-2058. 8. Winters, S.L. and Lee, D.R. Environ. Sci. Technol, 1987, 21, 1182-1186. 9. Widmer, S.K., Spalding, R.F. and Skopp, J. J. Environ. Qual. 1995, 24, 439-444. 10. Lapidus, L. and Amundson, N.R. J. Phys. Chem. 1952, 56, 984-988. 11. Hashimoto, I.; Deshpande, K.B.; Thomas, H.C. Ind Eng. Chem Fundamen. 1964, 3, 213-218. 12. Selim, H.M., Davidson, J.M. and Mansell, R.S. 1976,. p. 444-448. In Proc. Summer Comput. Simul. Conf., Wash DC. Simul Counc; La Jolla, CA. 13. van Genuchten, M. Th. and Wagenet R.J. Soil Sci. Soc. Am. J. 1989, 53, 1303-1310.



170 14. Nkedi-Kizza, P., Bigger, J.W., Selim, H.M., Van Genuchten, M.Th., Wierenga, P.J., Davidson, J.M. and Nielsen, D.R. Water Resour. Res. 1984, 20, 1123-1130. 15. Toride, N.; Leij, F.J.; van Genuchten, M. Th. The CXTFIT code for estimating transport parameters from laboratory or field tracer experiments, Version 2. Research Rpt. 137, U.S. Salinity Laboratory, USDA, ARS, Riverside, CA. 1995. 16. Gaber, H.M., Inskeep, W.P., Comfort, S.D. and Wraith, J.M. Soil Sci. Soc. Am. J. 1995, 59,160-67. 17. van Genuchten, M.Th. Res. Rep. no. 119. U.S. Salinity Lab., Riverside, CA. 1981a 18. Ogden, C.B., vanEs, H.M. and Schindelbeck. R.R. Soil Sci. Soc. Am. J. 1997, 61, 1041-1043. 19. Parker, J.C. and van Genuchten, M. Th. 1984, Virginia Agric. Exp. Stn. Bull. no. 84-3. 20. Gamerdinger, A.P, Lemley, A.T. and Wagenet, R.J. J. Environ. Qual. 1991, 20, 815-822. 21. Brusseau, M.L., Jessup, R.E. and Rao, P.S.C. Water Resources Research. 1989, 25, 1971-1988. 22. Rao, P.S.C., and Jessup, R.E.: 1983, In Chemical Mobility and Reactivity in Soil Systems, P. 183-201. American Society of Agronomy, Madison, WI. 23. Bahr, J.M., and Rubin, J. Water Resour. Res. 1987, 23, 438-452. 24. Khan, S.U. 1973, Can. J. Soil Sci. 53, 429-434. 25. Rao, P.S.C., Davidson, J.M. and Selim, H.M. Soil Sci. Soc. Am. J. 1979, 43, 22-28. 26. Brusseau, M.L., and Rao, P.S.C. Crit. Rev. Environ. Control. 1989, 19, 3399. 27. Brusseau, M.L. J. of Hydrol. 1998, 209, 147-165. 28. Brusseau, M.L., Jessup, R.E. and Rao, P.S.C. Environ. Sci. Technol. 1991, 25,134-142.


Modeling 2,4-D Transport in Turfgrass, Thatch and Soil - American

Recommend Documents